NVIDIA HairWorks is a cool and interesting piece of tech that already exists, but you probably never use it, because it absolutely kills performance (hi, Witcher 3). Still, development and progress continues, and here’s you can do with version 1.1 of the tech when you don’t have to worry about things like in-game framerates.
It looks, well, luxurious! This is 500,000 strands of individual hair flicking and tossing around like a bad shampoo commercial, in which NVIDIA is selling graphics cards containing herbal extracts and jojoba.
The clip isn’t technically an ad, though; NVIDIA is just promoting it. It’s the work of 3D artist Tarkan Sarim, who has worked for companies like WETA.
Here’s another video showing the hair swishing with some lighting effects added in.
Comments
9 responses to “The Future Of Video Game Hair: A Hilarious Shampoo Commercial”
I wonder what happened to AMD’s version. I can only recall Tomb Raider using it….
TresFX i think it was called. It didn’t seem to lower my FPS as much as hairworks did with Witcher 3.. (I’ve migrated gfx cards by then)
TresFX looked ok but playing through Tomb Raider again I’m seeing lots of really bad glitches with the hair in the semi-cutscenes. A lot of the time Lark’s hair will clip into her skull and reappear out of her forehead. HairWorks on the other hand seems to not have this issue. Still, let’s hope TresFX makes a full comeback, competition can only help here.
It’s still a thing, AMD are going to make it open source and free soon.
We are a while of these types of features in everyday games. PC being the first of course. Square Enix Tech Demo for DirectX 12 is a solid example. Some really revolutionary stuff that will change how games look, but its a long way off.
That demo was run on a computer with 4 top end Nvidia cards. Remember it was not even a full world, so more grunt is needed. But using the current trend its takes about 3 generations for 1 top end card to beat a top end sli config current gen… they have 4. Generations last 12 to 18 months (closer to 18 these days). Now SLI does not get 100% performance scaling from all cards, so doing some crazy napkin maths. We are still looking at about 7 to 8 years before a single high end graphics card can handle those scenes which currently comes with a price tag of 650 to 800 AUD. Consoles are even further away of course…
These hair sims look great, but for the average joe it costs to much to run vs how much grunt they have even in the PC world… And devs know this. Things like this will be apart of extra’s like gameworks, there for people with enough grunt currently or for people revisiting the game 3 years down the track. We are ages away from this being integrated normally into games and it will be a natural progression, more polygons, effects etc over the years. An example of this is in GTA 5 all the npc’s that walk on the sidewalk etc have the extract same poly count. That’s the type of depth that is involved, realizing how anal they are now with things like this also helps you appreciate how far away we really are.
TL:DR long long long way off (8 years or so) before we see Square Enix Tech Demo type stuff on even remotely affordable systems ie a 1500 dollar PC rig.
Decent hair effects in a game used to impress me….you know what would impress me more? Wet hair….amirite?
Nvidia have been optimizing it as of late. I think it will cost you 5-7fps now in Witcher 3 thought I haven’t played it for a while so I might be wrong. Originally it cost you 20fps…
I would like to see HairWorks used in Fallout4 on the animals/npc/etc.. F4 seems like it will have allot of free FPS to offer so it be nice.
Doing WET hair would require less processing as the hair clumps together requiring less strands, just a color tone and material change with added weight to clumped segments..
Witcher 3’s problem with Hairworks seems to have been more a misconfiguration than a performance problem in the library itself. For whatever reason CDPR left the tessellation rate at 64x, which is great for demonstrations like above, but excessive in a final game build. People who figured out how to reduce the tessellation rate found 8x and 16x only had minimal impact on framerate while giving almost indistinguishable visual quality.
What a missed chance for using a 3D model of John Romero.